Power and shift invariant detection of dynamically coupled networks (PSIICOS) from non-invasive MEG data

Abstract

MEG reflects electrical activity of neuronal assemblies that coalesce and decoalesce in time providing for massively parallel and dynamic flow of information exchange in the brain. There is a growing evidence that our behavior is mediated by simultaneous activity of several interacting and simultaneously active cortical networks. Not only the location of network nodes but also the temporal profiles of such interaction are of interest. The existing methods are primarily concerned with proposing novel measures of synchrony to be applied to the cortical region activity signals extracted from the MEG data using beamforming or other spatial filtering approaches.\n\nIn this work we propose to render the task of estimating network topology and temporal profiles as a source estimation problem but in the space of interacting topographies. The proposed point of view has a number of benefits as it allows to use the wealth of techniques and intuition accumulated in the community when dealing with source estimation problems. Operating in the interacting topographies space we first propose subspace projection method to significantly reduce the spatial leakage effect in the cross-spectral matrix. We then use subspace matching metrics to extract a set of networks with their synchronicity profiles that explain the variance in the sensor space cross-spectral matrix, much like regular dipoles and their activations explain the variance in the regular evoked responses.\n\nThe preliminary results of application of this method to a simple self-paced voluntary finger movement paradigm comparing allowed to observe functional coupling between primary motor and supplementary motor area and establish that this coupling can be characterized with near-zero phase delay between the oscillatory activity of these two regions. This observed synchrony pattern is physiologically plausible and is in agreement with the current hypothesis about the neural mechanics of the motor acts.